Is there a Python syntax to write/export a model as mps file using Highspy?

[hbshrestha]

The HIGHS documentation provides a simple example of building a model using highspy.
Is there an option to write or export mps file from the below given model using highspy?

import highspy
import numpy as np
h = highspy.Highs()

x0 = h.addVariable(lb = 0, ub = 4)
x1 = h.addVariable(lb = 1, ub = 7)

h.addConstr(5 <=   x0 + 2*x1 <= 15)
h.addConstr(6 <= 3*x0 + 2*x1)

h.minimize(x0 + x1)

[jajhall]

Yes, h.writeModel("foo.mps")

[hbshrestha]

Indeed, that's correct.

[jajhall]

I've tried to duplicate this behaviour, and I get the message

WARNING: There are empty or excessively-long column names: using constructed names with prefix "c"
WARNING: There are empty or excessively-long row names: using constructed names with prefix "r"

This is because there are no names associated with columns or rows. If the name isn't specified in the call to h.addVariable then, for the default value name: Optional[str] = None, no (blank) name is passed to HiGHS.

[rschwarz]

As a somewhat off-topic comment on this method: It seems that by default, it produces MPS in "fixed format" and HiGHS will truncate the coefficients to 10 digits of precision:

HiGHS/src/io/HMPSIO.cpp

Line 792 in 4a7f24a

fprintf(file, " %-8s %-8s %.10g\n", col_names[c_n].c_str(),

I've found that in the C++ method, there is a boolean flag where one can switch to a free format (with full precision, and also longer names for rows and columns), but the Python wrapper does not yet provide this choice, currently.

[jajhall]

Strange. My understanding is that MPS files are written out in free format with full precision, particularly if the names are longer than 8 characters, and regardless of calling language, with the fixed format reader only used if row/column names contain spaces. I'll check, nonetheless

[rschwarz]

Yes, I think this would be the best and safest default.